nvidia

/

Llama3-ChatQA-1.5-8B

like 555

Follow

NVIDIA 14.3k

Hi, thanks for sharing. I'm trying to reproduce the results from the original paper on the QuAC dataset using Llama-3-8B-instruct, but I'm only achieving an F1 score of 27.85 while the paper reports 33.60.

I'm wondering if there are any parameters (e.g., max_tokens) that may not be reflected in the Hugging Face that could affect this result.

Any guidance would be greatly appreciated!